What Is the New-Author Boost? How X’s Algorithm Elevates Posts From Low-Impression Authors

The New-Author Boost is a ranking multiplier that elevates posts toward the top of X’s home timeline when the author has accumulated fewer than 100 cumulative impressions, ensuring emerging creators gain visibility alongside established accounts.

The New-Author Boost operates within the xai-org/x-algorithm codebase as a fairness mechanism in the home-mixer pipeline. By detecting authors with low historical reach and applying a configurable score boost, the system prevents a "rich-get-richer" dynamic where only popular accounts dominate user feeds. This technical implementation examines author impression counts at query time and dynamically adjusts ranking scores before final placement.

How the New-Author Boost Works

The boost operates through a staged pipeline that evaluates author history and adjusts ranking position accordingly.

Impression Threshold Check

When the home-mixer evaluates a candidate post, it retrieves the author’s cumulative impression count from the query-level feature store. As defined in home-mixer/params/param.rs, the system compares this count against the ColdStartImpressionThreshold parameter (default: 100). Authors with impression counts below this threshold are classified as "new" and qualify for elevation.

Score Multiplication and Position Capping

For qualifying posts, the raw ranking score undergoes two sequential adjustments:

1. Multiplicative scaling – The raw score is multiplied by ColdStartImpressionScale (default: 2.0), effectively doubling the post’s competitiveness in the ranking algorithm.
2. Position constraint – The LowImpressionsMaxPositionRatio parameter (default: 0.20) prevents over-elevation by capping the boosted post’s position to the top 20% of the feed length. This ensures new authors gain visibility without completely displacing high-quality content from established creators.

The home-mixer/scorers/author_cold_start.rs file implements this logic, calculating a target_position based on the feed length multiplied by the maximum position ratio, then inserting the candidate at the higher of its raw or boosted position.

Configuration Parameters

The behavior of the New-Author Boost is controlled through three tunable parameters defined in the configuration layer:

// home-mixer/params/param.rs
param!(
    ColdStartImpressionThreshold,
    u64,
    "rust_home_mixer_cold_start_impression_threshold",
    100            // authors with < 100 impressions are “new”
);
param!(
    ColdStartImpressionScale,
    f64,
    "rust_home_mixer_cold_start_impression_scale",
    2.0            // boost factor applied to the raw score
);
param!(
    LowImpressionsMaxPositionRatio,
    f64,
    "rust_home_mixer_low_impressions_max_boost_position_ratio",
    0.20           // boosted posts cannot be placed beyond the top 20 % of the feed
);

These parameters allow X’s ranking engineers to tune sensitivity: lowering the threshold captures more authors, while increasing the scale factor provides stronger elevation to those who qualify.

Implementation in author_cold_start.rs

The scoring implementation retrieves these parameters at query time and applies conditional logic to each candidate post:

// home-mixer/scorers/author_cold_start.rs
let impression_threshold = query.params.get(ColdStartImpressionThreshold) as u64;
let impression_scale = query.params.get(ColdStartImpressionScale);
let max_pos_ratio = query.params.get(LowImpressionsMaxPositionRatio);

// For each candidate post:
if let Some(imp) = author_impression_count {
    if imp < impression_threshold {
        // Apply the boost
        let boosted_score = raw_score * impression_scale;
        // Ensure the boosted position stays within the allowed ratio
        let target_position = (feed_len as f64 * max_pos_ratio).floor() as usize;
        // …insert the candidate at `target_position` in the ranking list…
    }
}

This code path executes during the home timeline’s scoring phase, ensuring that elevation decisions occur before final ranking consolidation.

Telemetry and Monitoring

To ensure the New-Author Boost functions as intended without degrading feed quality, X’s pipeline logs all boosted posts to the telemetry metric home_mixer.cold_started_posts_total. This monitoring allows engineers to track how frequently the boost triggers, measure engagement rates on elevated content, and adjust the impression threshold or scale factors based on real-world performance data.

Summary

• The New-Author Boost targets authors with fewer than 100 cumulative impressions, preventing established accounts from monopolizing the home timeline.
• Elevation occurs through multiplicative score scaling (default 2.0×) combined with a position cap (top 20% of feed) to balance visibility and quality.
• Configuration resides in home-mixer/params/param.rs, while the scoring logic lives in home-mixer/scorers/author_cold_start.rs.
• Boosted posts are tracked via the home_mixer.cold_started_posts_total telemetry metric for ongoing optimization.

Frequently Asked Questions

What is the New-Author Boost in X’s algorithm?

The New-Author Boost is a ranking adjustment mechanism that identifies posts from creators with low historical visibility (fewer than 100 impressions) and elevates their position in the home timeline. It applies a multiplicative score boost and position cap to ensure emerging voices receive algorithmic promotion without overwhelming the feed.

How does the algorithm prevent new-author posts from completely dominating the feed?

The system enforces the LowImpressionsMaxPositionRatio parameter (default 0.20), which restricts boosted posts to the top 20% of the timeline regardless of their raw score. This cap ensures that while new authors gain visibility, they cannot displace all high-quality content from established creators.

Where can I find the source code for the New-Author Boost?

The implementation resides in two primary files within the xai-org/x-algorithm repository: home-mixer/scorers/author_cold_start.rs contains the scoring logic, while home-mixer/params/param.rs defines the configurable parameters including the impression threshold and scale factor.

How does X measure the effectiveness of the New-Author Boost?

All posts receiving the boost are logged to the home_mixer.cold_started_posts_total telemetry metric. Engineers monitor this data alongside engagement metrics to ensure the boost surfaces quality content from new creators rather than simply promoting low-value posts, allowing for data-driven adjustments to the threshold and scale parameters.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}